Question

我试图了解Caret包中的5折交叉验证算法，但我无法找到如何获得每个折叠的列车集和测试集，我也无法从类似的建议问题中找到它。想象一下，如果我想通过随机森林方法进行交叉验证，我会执行以下操作：

set.seed(12)
train_control <- trainControl(method="cv", number=5,savePredictions = TRUE)
rfmodel <- train(Species~., data=iris, trControl=train_control, method="rf")
first_holdout <- subset(rfmodel$pred, Resample == "Fold1")
str(first_holdout)
'data.frame':   90 obs. of  5 variables:
$ pred    : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1     
$ obs     : Factor w/ 3 levels "setosa","versicolor",..: 1 1 1 1 1 1 1 1 1 1 
$ rowIndex: int  2 3 9 11 25 29 35 36 41 50 ...
$ mtry    : num  2 2 2 2 2 2 2 2 2 2 ...
$ Resample: chr  "Fold1" "Fold1" "Fold1" "Fold1" ...

Fold1中的这90个观察值是否用作训练集？如果是，那么此折叠的测试集在哪里？

Answer 1

 str(rfmodel)

执行的模型以下面的形式存储所有内容。下面的control存储了index和indexOut中用于培训和相应保留的样本的索引。

 names(rfmodel)
 #  [1] "method"       "modelInfo"    "modelType"    "results"      "pred"        
 #  [6] "bestTune"     "call"         "dots"         "metric"       "control"     
 # [11] "finalModel"   "preProcess"   "trainingData" "resample"     "resampledCM" 
 # [16] "perfNames"    "maximize"     "yLimits"      "times"        "levels"      
 # [21] "terms"        "coefnames"    "xlevels"

培训和保留样本的索引路径

 # Indexes of Hold Out Sets
 rfmodel$control$indexOut

 # Indexes of Train Sets for above hold outs
 rfmodel$control$index

Caret交叉验证中每个折叠的测试集和训练集

1 个答案: